Introduction
The treatment landscape for relapsed/refractory (R/R) non-Hodgkin lymphoma (NHL) has been fundamentally transformed by two immunotherapy platforms: chimeric antigen receptor T-cell (CAR-T) therapy and CD20×CD3 bispecific antibodies (BsAbs). Although both modalities harness T-cell cytotoxicity against malignant B cells, they differ profoundly in mechanism of delivery, manufacturing requirements, toxicity burden, and practical accessibility. This review provides a synthesis of current clinical evidence across three key NHL subtypes—diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), and mantle cell lymphoma (MCL)—to guide oncologists and hematologists in individualizing therapy selection.
Efficacy by NHL Subtype
Diffuse Large B-Cell Lymphoma
In second-line (2L) transplant-eligible DLBCL, CAR-T therapy has established superiority over standard salvage chemoimmunotherapy in two pivotal randomized trials. The ZUMA-7 trial of axicabtagene ciloleucel (axi-cel) demonstrated a median event-free survival (EFS) of 8.3 months versus 2.0 months (hazard ratio [HR] 0.398), with an objective response rate (ORR)/complete response (CR) of 83%/65%, and a 48-month overall survival (OS) of 54.6% versus 46.0% for standard of care 33. The TRANSFORM trial of lisocabtagene maraleucel (liso-cel) similarly showed a median EFS of 29.5 versus 2.4 months (HR 0.375), with ORR/CR of 87%/74% 33. Notably, tisagenlecleucel (tisa-cel) did not demonstrate EFS benefit over standard therapy in the BELINDA trial, underscoring sensitivity to vein-to-vein time and bridging strategy 33.
In later-line R/R LBCL, both modalities are active. The five-year ZUMA-1 data for axi-cel show a 5-year EFS of 34% overall and 57% among complete responders, with median OS of 25.8 months 33. For BsAbs in this setting, the STARGLO randomized trial of glofitamab plus gemcitabine/oxaliplatin (GemOx) demonstrated OS benefit over R-GemOx (HR 0.60; median not reached vs. 13.5 months) and progression-free survival (PFS) of 13.8 versus 3.6 months (HR 0.41), with CR rates of 58.5% versus 25.3% 33. Odronextamab monotherapy in R/R DLBCL (ELM-2 trial; n=127) achieved an ORR of 52.0% and CR of 31.5%, with a median duration of response (DOR) of 10.2 months and median OS of 9.2 months 3.
Follicular Lymphoma
CAR-T therapy achieves high response rates and durable remissions in R/R FL. In the 5-year follow-up of ZUMA-5, axi-cel showed high best overall response and complete response rates, with particularly strong outcomes in the FL subgroup 33. Liso-cel in R/R FL yielded ORR ≥96% with 12-month DOR of 89.8% and PFS of 91.3% in the 2L cohort 33.
BsAbs also achieve high response rates in FL. Mosunetuzumab in a pivotal phase II cohort (n=90) produced an ORR of 80% and CR of 60%, with a median PFS of 24 months and median OS not reached at 36-month follow-up 2. Epcoritamab in the EPCORE NHL-1 FL cohort (n=128) demonstrated an ORR of 82.0% and CR of 62.5%, with median PFS of 15.4 months; among patients treated for ≥10 cycles, 92% maintained response 2. Odronextamab in the ELM-2 FL cohort reported an ORR of 80% and CR of 73%, with a 12-month CR maintenance probability of 75% and median PFS of 20.7 months 2. Measurable residual disease (MRD) clearance has been associated with favorable outcomes in some FL studies, but its role as a consistent predictor of prolonged PFS across all FL cohorts should be interpreted cautiously 23.
Mantle Cell Lymphoma
CAR-T therapy with brexucabtagene autoleucel (KTE-X19; Tecartus) in the ZUMA-2 trial demonstrated ORR of ~91% and CR of 64–73%, with 5-year OS of 39–54% in prior BTK inhibitor (Bruton's tyrosine kinase inhibitor)–treated cohorts 33. BsAb-specific data in MCL remain sparse in the retrieved literature; however, glofitamab was reported to produce objective responses and stimulate circulating CAR-T cell expansion in two MCL patients relapsing after CAR-T therapy 24. Novel non-covalent BTK inhibitors and BsAbs are under active investigation in MCL 9.
Toxicity Profiles
The table below summarizes key toxicity comparisons across modalities.
| Toxicity Domain | CAR-T (axi-cel/liso-cel/brexu-cel) | Bispecific Antibodies |
|---|---|---|
| Grade ≥3 CRS | 6–17% (axi-cel); 1.1–2.2% (liso-cel) 34 | 1.6–7.5%; reduced with step-up dosing 13 |
| Grade ≥3 ICANS | 21–35% (axi-cel); 4–4.3% (liso-cel) 34 | Rare (0–6.4%); mostly grade 1–2 113 |
| Any-grade CRS | 60% (real-world registry) 35 | 44–66% FL; 53–56% DLBCL 23 |
| Grade ≥3 Neutropenia | Common; prolonged cytopenias in ~22% 33 | 26% (odronextamab DLBCL) 3 |
| Grade ≥3 Infections | Significant; COVID-19 mortality noted 2 | 38.6% (odronextamab); 37% (post-CAR-T BsAb) 314 |
| ICANS onset | Median day 5 post-infusion 35 | Rare; no ICANS in odronextamab optimized cohort 3 |
| Treatment-related mortality | 0–5.7% 34 | 3.9% (odronextamab DLBCL) 3 |
Cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) represent shared toxicities, but with quantitatively distinct profiles. Real-world 2024–2025 registry data across 475 patients receiving anti-CD19 CAR-T showed that new-onset CRS peaked at day 3 post-infusion with no new cases after day 14; ICANS peaked at day 5, with only one case (0.7%) beyond day 14 35. CAR-T carries meaningfully higher ICANS risk—particularly with axi-cel, where grade ≥3 neurologic events reached 21% in the 2L DLBCL setting—compared to BsAbs, where grade ≥3 ICANS is rare across pivotal trials 3334.
Anti-BCMA CAR-T products carry an additional spectrum of delayed neurotoxic syndromes, including IEC-associated Parkinsonism and cranial nerve palsy, occurring weeks to months after infusion—a pattern not reported with CD20×CD3 BsAbs 38. Late-onset hypogammaglobulinemia and B-cell aplasia affect 47–73% of anti-CD19 CAR-T responders at one year, mandating long-term immunoglobulin monitoring and replacement in selected patients 35.
Infections represent a major cross-modality concern. COVID-19–attributable mortality was substantial in BsAb FL trials conducted during the pandemic (6 of 128 patients in EPCORE NHL-1; 8 of 128 in ELM-2 FL) 2. Grade ≥3 infections occurred in 38.6% of odronextamab-treated DLBCL patients 3 and 13% in real-world post-CAR-T BsAb recipients 14.
Toxicity Management Strategies
Step-up dosing is the cornerstone of BsAb CRS mitigation, titrating from subtherapeutic priming doses to full therapeutic doses over the initial treatment cycles. Odronextamab's optimized three-step protocol (0.7–4–20 mg) eliminated grade 4–5 CRS 3. Glofitamab employs obinutuzumab 1,000 mg pretreatment one week prior, reducing post-dose IL-6 levels and enabling safer dose initiation 1. Epcoritamab's subcutaneous administration with delayed peak serum concentrations further attenuates cytokine release relative to intravenous dosing 1.
Corticosteroid prophylaxis protocols differ: mosunetuzumab requires dexamethasone 20 mg for 4 doses (cycles 1–2); epcoritamab requires prednisolone 100 mg for 16 doses; odronextamab requires dexamethasone 20 mg for 15 doses 2. The cumulative corticosteroid burden is substantially higher with epcoritamab and odronextamab protocols.
For CAR-T toxicity management, tocilizumab (anti-IL-6 receptor) is first-line for CRS, with retrospective data confirming that tocilizumab does not impair CAR-T efficacy or OS 34. Corticosteroids (dexamethasone or methylprednisolone) are concurrent first-line agents for ICANS and high-grade CRS. Refractory cases may warrant anakinra, ruxolitinib, siltuximab, or antithymocyte globulin 34. Emerging evidence from ZUMA-24 supports outpatient axi-cel administration with prophylactic steroids achieving 95% ORR without grade ≥3 CRS 33. Real-world data suggest that flexible monitoring beyond day 14 is safe, given the absence of new-onset CRS or ICANS after this point in the vast majority of patients—findings that challenge the current 4-week FDA REMS proximity requirement and may improve patient access and equity 35.
Practical and Clinical Positioning
Manufacturing timelines remain the most consequential practical distinction. CAR-T requires 3–6 weeks from leukapheresis to infusion, during which bridging therapy may be necessary for rapidly progressive disease. BELINDA highlighted how extended vein-to-vein intervals can compromise outcomes in aggressive DLBCL 33. BsAbs, as off-the-shelf products, can be initiated within days of treatment decision, a critical advantage for patients with rapidly evolving disease 337.
Sequencing considerations are increasingly supported by data. BsAbs retain meaningful activity after CAR-T failure: in a multicenter real-world analysis (n=92, predominantly glofitamab), the ORR post-CAR-T was 43%, with outcomes stratified sharply by time to CAR-T relapse—ORR was 60% in late relapse (>6 months) versus 29% in early relapse (≤3 months) 14. Notably, patients receiving BsAbs as first salvage after CAR-T failure showed superior median PFS (not reached) and OS (not reached) compared to those receiving intervening therapy before BsAb 14. Conversely, epcoritamab showed an ORR of 54.1% even in CAR-T–refractory patients 13.
Route of administration and setting: Epcoritamab's subcutaneous formulation allows predominantly outpatient initiation; in a prospective outpatient cohort, 90% of patients received their first full dose as outpatients 33. Mosunetuzumab offers a defined-duration regimen: patients with a complete response after 8 cycles discontinue therapy, whereas patients with a partial response or stable disease may continue treatment for up to 17 cycles unless disease progression or unacceptable toxicity occurs. Odronextamab and epcoritamab continue until progression or intolerance 2.
Patient eligibility: CAR-T is often limited in elderly or heavily comorbid patients by fitness requirements, lymphodepletion, and inpatient monitoring mandates. BsAbs—including epcoritamab combined with R-mini-CHOP—have demonstrated high CR rates and MRD negativity in very elderly first-line LBCL patients, extending therapeutic reach to populations ineligible for CAR-T 33.
Comparative Synthesis and Conclusions
CAR-T cell therapy delivers unmatched depth and durability—particularly in 2L transplant-eligible DLBCL (axi-cel, liso-cel) and R/R FL—and represents curative-intent therapy in eligible patients. However, its ICANS burden, manufacturing latency, and complex infrastructure requirements constrain its universal applicability. BsAbs offer scalable, rapid, off-the-shelf immunotherapy with manageable predominantly low-grade CRS, outpatient feasibility, and meaningful activity across all three major B-cell NHL subtypes, including in post-CAR-T settings. The STARGLO trial's demonstration of OS benefit for glofitamab plus GemOx in later-line DLBCL marks a critical milestone for BsAb-based regimens 33.
Critically, cross-trial efficacy comparisons must be interpreted with caution given heterogeneity in patient populations, prior lines of therapy, assessment timing, and trial design 33. Direct head-to-head randomized comparisons remain absent. Nonetheless, an evolving framework positions CAR-T as the preferred modality where manufacturing is feasible and patients are fit—particularly in 2L DLBCL and early-relapse FL—while BsAbs serve as the optimal choice when CAR-T is inaccessible, contraindicated, or when rapid intervention is required. Ongoing trials evaluating BsAbs in earlier treatment lines, in combination with chemoimmunotherapy, and in first-line elderly populations, alongside next-generation allogeneic and point-of-care CAR-T platforms, will continue to refine this evolving therapeutic landscape 283337.